Electron Transfer Methods in Open Systems | The Journal of Physical

Jul 30, 2019 - Electron Transfer Methods in Open Systems ... and fourth order can be obtained as a particular case of zero- and second-order Green's ...
0 downloads 0 Views 820KB Size
Subscriber access provided by KEAN UNIV

Article

Electron Transfer Methods in Open Systems Nicolas Bergmann, and Michael Galperin J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.9b05813 • Publication Date (Web): 30 Jul 2019 Downloaded from pubs.acs.org on July 31, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Electron Transfer Methods in Open Systems Nicolas Bergmann† and Michael Galperin∗,‡ †Department of Chemistry, Technical University of Munich, D-85748 Garching, Germany ‡Department of Chemistry & Biochemistry, University of California San Diego, La Jolla, CA 92093, USA E-mail: [email protected] Phone: +1 858 246 0511

1

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract Utilization of electron transfer methods for description of quantum transport is popular due to simplicity of the formulation and its ability to account for basic physics of electron exchange between system and baths. At the same time, necessity to go beyond simple golden rule-type expressions for rates was indicated in the literature and ad hoc formulations were proposed. Similarly, kinetic schemes for quantum transport beyond usual second order Lindblad/Redfield considerations were discussed. Here we utilize recently introduced by us nonequilibrium Hubbard Green’s functions diagrammatic technique to analyze construction of rates in open systems. We show that previous considerations for rates of second and fourth order can be obtained as a particular case of zero and second order Green’s function diagrammatic series with bare diagrams. We discuss limitations of previous considerations, stress advantages of the Hubbard Green’s function approach in constructing the rates and indicate that standard dressing of the diagrams is a natural way to account for additional baths/degrees of freedom when formulating generalized expressions for the rates.

2

ACS Paragon Plus Environment

Page 2 of 28

Page 3 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Introduction Electron transfer processes are the heart of oxidation-reduction reactions which play important role in chemistry and biology. 1–3 Theoretical description of electron transfer rates at the level of the Marcus theory 4–10 is widely utilized for the description of variety of phenomena from photovoltaics, batteries design and catalysis in chemistry 11–16 to photosynthesis, vision and sense of smell in biology. 17–23 Interfacial electron transfer is behind many vital biological processes. 24–28 Recently, biomolecules were utilized as building blocks in electric circuits. Biomolecular junctions are useful as a tool to study properties of molecules and as potential bioelectronic devices. For example, electron transport was measured through DNA, 29–31 oligopeptides 32,33 and electron transfer proteins. 34–36 Also, STM junctions were suggested as a convenient tool for DNA detection and sequencing. 37 Traditionally, electron transfer theory considers isolated donor-bridge-acceptor systems. Nevertheless, this theoretical approach appears to be useful also in description of electron transport in open molecular systems such as, e.g., redox molecular junctions. 38–41 Formal relationship between electron transfer rates and molecular conduction was discussed in the literature. 42 In both intra-system and interfacial processes in condensed phase electron transfer may be assisted by intermediates. For example, intermediate states play important role in photovoltaic 43 or long-range DNA electron transfer processes. 44–46 Effects of intermediates on electron transfer are often discussed in terms of super-exchange vs. hopping mechanism. Clearly, the two mechanisms are limiting cases (coherent and completely incoherent) of the same process, and attempts to unify electron transfer rate expression were done in the literature. 47,48 However, these attempts are ad hoc perturbation theory considerations, and an ordered way of rate simulations is still lacking. Constructing generalized expressions for rates is also at the heart of kinetic equations approaches to transport. For example, Refs. 49,50 are similar in the spirit effort to introduce 3

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

a scheme for constructing generalized expressions for rates. Advantage of this approach is possibility to account for higher order processes in system-bath couplings in an ordered way of bare perturbation theory. Still, questions of, e.g., accounting for additional degrees of freedom (such as other baths) within a particular order in the system-primary baths coupling remains open. Also, the methodology has usual for quantum master equations restriction of applicability only in the high temperature regime (thermal energy should be big relative to characteristic energy of the system-bath coupling). We note in passing that in interacting open systems kinetic schemes should be applied with caution. 51 Nonequilibrium Green’s functions (NEGF) 52 is a tool capable to describe both superexchange and hopping transport regimes, as well as smooth transition between the two limiting cases. Diagrammatic perturbation theory is a way to account for interaction with intermediate (or additional) degrees of freedom in an ordered form. Also, Green’s function considerations are applicable in any temperature regime. However, identifying rates is not possible within the standard NEGF. Here, we show that recently introduced by us many-body NEGF flavor, the Hubbard NEGF, 53 while retaining advantages of the Green’s function methods is capable to provide connection with kinetic schemes. So that ordered construction of generalized rates becomes possible employing the Hubbard Green’s function diagrammatic technique. Structure of the paper is the following. After introducing junction model we give a short overview of the Hubbard NEGF introducing conceptual details not presented in its original introduction in Ref. 53. After this we present connection between the Hubbard NEGF and kinetic schemes of Refs. 49,50 showing a way to formulate rates within the Hubbard NEGF. We conclude by discussing advantages of Green’s functions formulation and outlining directions for future research.

4

ACS Paragon Plus Environment

Page 4 of 28

Page 5 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Theory Model. We consider generic model of a junction which consists of a molecule, M , coupled to two contacts, L and R. Depending on particular problem, molecular (system) part can describe electronic, vibrational, optically-dressed (e.g. polariton) or any other degrees of freedom. We assume that quantum chemistry problem for the isolated system has been solved, and many-body eigenstates |Si and their energies ES are known. We note in passing that even in systems with very big (or even infinite) number of many-body eigenstates, energetics of the junction (bias, driving laser field, etc.) allows to identify a finite subset, which is enough for first principles simulation of experimental data for realistic systems. 54–58 Contacts (baths) are assumed to be reservoirs of free charge carriers each at its own equilibrium. Second quantization is utilized to treat baths’ degrees of freedom. This is minimal model for discussion of electron transfer rates within the Hubbard NEGF. Additional baths and/or degrees of freedom (e.g., phonons) can be added to the consideration in a straightforward manner. Hamiltonian of the minimal model is

ˆ =H ˆM + H

 X  ˆ ˆ HK + VM K

(1)

K=L,R

ˆM = H

X

|SiES hS|

(2)

εk cˆ†k cˆk

(3)

S∈M

ˆK = H

X k∈K

VˆM K =

X X

Vk,S1 S2 cˆ†k

 |S1 ihS2 | + H.c.

(4)

S1 ,S2 ∈M k∈K

ˆ M and H ˆ K (K = L, R) are molecule and contacts Hamiltonians, and VˆM K introduces Here H system-baths coupling. cˆ†k (ˆ ck ) creates (annihilates) electron in single-electron state k in the contacts. Vk,S1 S2 is matrix element for electron transfer from system to bath in which system goes from state |S2 i to |S1 i.

5

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 28

Below we will be interested in electron flux through the junction and corresponding intrasystem electron transfer rates. Note that while we focus on charge current, the consideration is more general and can equivalently be applied to calculation of any other intra-system rates (e.g., related to photon or energy flux) or to multi-time correlation functions as employed in, e.g., nonlinear optical spectroscopy. 59 Hubbard NEGF. Central object of interest is the single particle Hubbard Green’s function, which is defined on the Keldysh contour 52,60 as ˆ S1 S2 (τ ) X ˆ † (τ 0 )i GS1 S2 ,S3 S4 (τ, τ 0 ) = −ihTc X S3 S4

(5)

Here XS1 S2 ≡ |S1 ihS2 | is the Hubbard (or projection) operator, Tc is the contour ordering operator, and τ and τ 0 are the contour variables. Advantage of the Hubbard over standard NEGF for our study is possibility to access information on many-body states of the system. As we show below, this moment is crucial for formulating general expressions for rates within the Green’s function methodology. Historically, Hubbard Green’s functions where introduced for treatment of strongly correlated extended (lattice type) equilibrium systems in Ref. 61. Diagrammatic technique (expansion around atomic limit) for such equilibrium Green’s functions 62,63 is based on assumption of equilibrium character of the uncoupled system’s density operator

ρˆ0 =

1 −Hˆ 0 /kB T e ; Z0

ˆ

Z0 = Tr e−H0 /kB T

(6)

At nonequilibrium, Hubbard Green’s functions where used for transport simulations employing relations derived from equation-of-motion considerations and functional derivatives in auxiliary fields. 64–69 While the latter approach is very useful, it lacks rigor of ordered diagrammatic expansion and provides only vague rules about choice of auxiliary fields and terms resulting from performing functional derivatives. Recently, we introduced nonequilibrium version of the Hubbard diagrammatic tech6

ACS Paragon Plus Environment

Page 7 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

nique 53 making it applicable to nonequilibrium impurity-type (molecular junction) problems. Contrary to original lattice type formulation, we introduce baths and utilize state representation only for the system, while bath degrees of freedom are treated within standard second quantization. Thus, perturbative expansion in the system-baths couplings in each order yields product of two multi-time correlation functions: one for the system and one for the bath operators (see Ref. 53 for details). Because baths are assumed to be non-interacting, the latter can be treated using the standard Wick’s theorem. 70,71 To evaluate multi-time correlation function of Hubbard (system) operators we employ usual for NEGF assumption of steady-state being independent of initial condition at infinite past. Thus, assuming equilibrium initial system state, Eq. (6), correlation function of Hubbard operators is evaluated using diagrammatic technique of Refs. 62,63. As a result, Hubbard NEGF appears to be a modified (by presence of baths and nonequilibrium character of the system) version of the lattice diagrammatic technique (see Ref. 53 for details). The technique appears to be quite stable over wide range of parameters, 72 helpful in evaluation of electronic friction in junctions, 73,74 and useful as a convenient tool in first principles simulations of optoelectronic devices. 56–58 It is important to realize, however, that requirement of equilibrium character of the uncoupled system density operator, Eq. (6), in principle can be relaxed. Indeed, diagrammatic technique of Refs. 62,63 is based on commutation properties of the Hubbard operators (interaction representation) ˆ S1 S2 (t) ˆ S1 S2 (t) X ˆ † (t0 ) = e−i(ES2 −ES1 )(t−t0 ) δS2 ,S4 X ˆ S1 S3 (t0 ) ± X ˆ † (t0 ) X X S3 S4 S3 S4

(7)

with sign chosen according to the operators statistics (Bose or Fermi - see Ref. 53 for details) and ability to interchange Hubbard and equilibrium density operators

ˆ S1 S2 ρˆ0 = ρˆ0 X ˆ S1 S2 e(ES1 −ES2 )/kB T X

7

ACS Paragon Plus Environment

(8)

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 28

Thus, it is clear that diagrammatic technique for the Hubbard operators can be equivalently formulated for any form of system’s density operator as long as the latter is a function of molecular (system) Hamiltonian only

ˆM ) ≡ ρˆ0 = f (H

X

|Si f (ES ) hS|

(9)

S

While this observation does not change numerical procedure for the Hubbard NEGF, it yields two important conceptual consequences. First, the Hubbard NEGF can be considered as a natural tool for expansion around results of a quasi-particle-type consideration performed in the basis of many-body states capable to introduce states broadening and bath-induced coherences, which were missed in the latter. For example, such quasi-particle-type consideration is the Markov Redfield/Lindblad quantum master equation. In this case f (ES ) in (9) is probability PS of state |Si to be observed. Second, considering situation where only one state is populated, PS = 1, nonequilibrium diagrammatic technique for Hubbard Green’s functions provides access to traditional expressions for rates of transitions from state |Si to all other states of the system. Below we discuss details of this Green’s function-to-kinetic scheme connection.

Results We now turn to discuss how Hubbard NEGF can be used to define generalized expressions for transfer rates. We explore connection to the kinetic scheme presented in Refs. 49,50, thus identifying rate expressions in terms of the Hubbard NEGF, and indicate how the rates expressions can be generalized. We start by considering equation-of-motion for the probability of state |Si to be observed (Heisenberg picture) ˆ SS (t)i PS (t) ≡ hX

8

ACS Paragon Plus Environment

(10)

Page 9 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

ˆ of (1). Writing Heisenberg equation of motion for (10) under driving by the Hamiltonian H and using standard NEGF derivation (similar to the derivation leading to the celebrated Meir-Wingreen expression for current; 52 details of the derivation are given in the Supporting Information) one gets X Z t d dt0 PS (t) = 2Re dt S1 ,S2 ,S3 −∞  > 0 < 0 > 0 σSS (t − t0 ) G< S1 S2 ,SS3 (t − t) − σSS3 ,S1 S2 (t − t ) GS1 S2 ,SS3 (t − t) 3 ,S1 S2 −

σS>3 S,S1 S2 (t

0

−t

0 ) G< S1 S2 ,S3 S (t

− t) +

σS S1 S2 ,S3 S (t

(11)

 − t)

Here G) is the lesser (greater) projection 52 of the Hubbard Green function (5) and σ ) is the lesser (greater) projection of the self-energy due to coupling to contacts L and R X X

σS1 S2 ,S3 S4 (τ, τ 0 ) =

VS1 S2 ,k gk (τ, τ 0 ) Vk,S3 S4 ≡

X

σSK1 S2 ,S3 S4 (τ, τ 0 )

(12)

K=L,R

K=L,R k∈K

where gk (τ, τ 0 ) = −hTc cˆk (τ ) cˆ†k (τ 0 )i is the Green function of free electron in single-particle state k. Expression (11) is exact, and our goal is to represent it in the form of rate equation, which will allow to identify expressions for the rates. Following Refs. 49,50 we will be interested in rates of second and fourth order in the system-baths couplings. Taking into account that Hubbard NEGF diagrammatic technique expands in system-baths coupling and noting that second order in the coupling already enters (11) via self-energy (12), it is natural to expect that second order rates should result from zero order of the Hubbard GFs expansion, while fourth order rates should be accessible from second order of the Hubbard NEGF diagrammatic series. Figure 1 shows diagrams of zero (panel a) and second (panels b-j) order expansion in the system-bath coupling. Directed straight and wavy lines represent Fermion propagator gm (τ, τ 0 ) and self-energy σmm0 (τ, τ 0 ), respectively. Fermion propagator gm (τ, τ 0 ) is one of main 9

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1: Diagrams of the Hubbard NEGF (5) expansion in system-baths coupling. Shown are (a) zero and (b-j) second order diagrams. See text for details.

10

ACS Paragon Plus Environment

Page 10 of 28

Page 11 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

constituents of the nonequilibrium diagrammatic technique for Hubbard Green’s functions. It represents propagation of quasiparticle; Hubbard Green’s function in addition has also information on quasiparticle spectral weight (see Ref. 53 for details of the Hubbard NEGF diagrammatic technique). Self-energy σmm0 (τ, τ 0 ) yields effect of contacts (bath) on the molecule (system). It is defined in Eq. 12. Here m ≡ S1 S2 is single electron transition between pair of many-body states |S1 i and |S2 i, i.e. NS1 +1 = NS2 (NS is number of electrons in state |Si). Dashed line stands for Boson propagator gb (τ, τ 0 ) in the same charging block, i.e. b ≡ S1 S2 with NS1 = NS2 . Directed double line represents two-electron propagator db (τ, τ 0 ), where b ≡ S1 S2 with NS1 + 2 = NS2 . Empty circle stands for (zero-order) spectral weight hFˆm1 m2 (τ )i0 ≡



ˆ m1 (τ ); X ˆ † (τ ) X m2 0

(13)

and oval with two circles is correlation function

hδ Fˆm1 m2 (τ ) δ Fˆm3 m4 (τ 0 )i0

(14)

where δ Fˆmm0 = Fˆmm0 − hFˆmm0 i0 . Filled circles stand for ‘pruned’ vertices. 63 Finally, red dashed line indicates the end point of the diagram. For more details see Ref. 53. We note that Fig. 1 presents bare diagrams. As discussed in the previous section, the Hubbard NEGF may be considered as expansion around results of the Markov Redfield/Lindblad quantum master equation. Assuming we are dealing with such an expansion, each diagram can be easily represented in terms of state probabilities. For example, lesser and greater projections of the Fermion propagator gm (τ, τ 0 ) are (m ≡ S1 S2 with NS1 + 1 = NS2 ) PS2 0 e−i(ES2 −ES1 )(t−t ) PS1 + PS2 PS1 0 > (t − t0 ) = −i gm e−i(ES2 −ES1 )(t−t ) PS1 + PS2

< gm (t − t0 ) = i

11

ACS Paragon Plus Environment

(15) (16)

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 28

while its casual and anti-casual projections are

c > < gm (t − t0 ) = θ(t − t0 )gm (t − t0 ) + θ(t0 − t)gm (t − t0 )

(17)

c˜ < > gm (t − t0 ) = θ(t − t0 )gm (t − t0 ) + θ(t0 − t)gm (t − t0 )

(18)

Similarly one can evaluate other elements of the diagrams, and represent the diagrams in terms of states probabilities PS and time-dependent factors. We now turn to order-by-order analysis of the diagrams. Before discussing the contributions we have to stress difference in Green function (Hilbert space) and quantum master equation (Liouville space) languages mentioned also in our previous publication: 75,76 time arrangements on the Keldysh contour called diagrams in the QME language are projections in the language of GFs. The difference is of minor importance for the zero-order contribution, because the latter has only one diagram (see Fig. 1a). However, as is discussed later, it becomes critical for understanding higher order contributions and connection between the methods. Zero-order contributions. Zero-order contribution to the Hubbard GF (5) is (0) Gmm0 (τ, τ 0 ) = gm (τ, τ 0 )hFˆmm0 (τ 0 )i0

(19)

Here m = S1 S2 and m0 = S3 S4 with NS1 + 1 = NS2 and NS3 + 1 = NS4 . We note that there are four different projections of (19) contributing to (11) - one projection for each one of the terms on the right side of the expression. The projections are given in Fig. 2. Note there are eight projections when considering also complex conjugates taken into account by 2 Re . . . in (11) - those are obtained by interchanging time positions between contour branches and flipping arrows. Taking projections of Eq. (19) on the Keldysh contour, substituting zero-order expressions for the locator and spectral weight, Eqs. (15)-(18), and utilizing the results in Eq. (11) yields expressions for the second order transfer rates from initial state |Si i to final state |Sf i (see 12

ACS Paragon Plus Environment

Page 13 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 2: Projections of the zero-order contribution, diagram in Fig. 1a, to equation-ofmotion (11): the four projections correspond to the four terms in the right side of the expression.

13

ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 28

Supporting Information for details) (2)

WSf ←Si = i σS>f Si ,Sf Si (ESi − ESf ) − i σS